RTL8814AU: match TX descriptor to kernel-driver field-for-field#28
Closed
josephnef wants to merge 1 commit into
Closed
RTL8814AU: match TX descriptor to kernel-driver field-for-field#28josephnef wants to merge 1 commit into
josephnef wants to merge 1 commit into
Conversation
Usbmon capture of the working aircrack-ng/morrownr kernel driver
injecting a probe-request on the 8814 (channel 6, monitor mode) was
diffed against devourer's TX descriptor, both 140-byte single-
fragment frames. The first 32-byte descriptor block differed in
seven fields. Update devourer's send_packet so the descriptor is
byte-identical to the kernel-driver's:
MACID = 1 (was 0) broadcast/default CAM
RATE_ID (non-VHT) = 8 (was 7) rate-table index
GID = 63 (was 0) no-group default
SW_DEFINE = 1 (was 0) DriverFixedRate flag
RETRY_LIMIT_ENABLE = 1 (was 0) mgmt-frame default
DATA_RETRY_LIMIT = 12 (was 0) from rtl8814au_xmit.c:267
SPE_RPT = 0 (was 1) kernel does not set
DISABLE_FB = 0 (was 1) kernel does not set
The previous values came from earlier WIP comments speculating about
chip behavior. Live usbmon diff against the kernel driver shows the
actual field set.
Verified on CF-938AC (0bda:8813, channel 6):
- Devourer's first TX bulk-OUT now reads:
64002885 01120800 0000003f 00010000 00003200 00000000 \
01000000 76a90000
Byte-identical to kernel-driver's TX descriptor.
- With chip primed by the kernel driver (insmod 8814au.ko, set
monitor, set ch 6, rmmod) and devourer started with
DEVOURER_SKIP_INIT=1: 781 of 781 bulk-OUT URBs complete with
status=0 (chip ACKs every URB). With the previous descriptor,
URBs completed status=0 too but on-air TX never happened — the
chip silently dropped frames whose descriptor field values it
didn't accept.
- 8814 RX still works end-to-end (10+ packets in demo window).
Note: this does NOT make devourer's full-init TX work end-to-end.
Devourer's HAL init still leaves the chip in a state where EP 0x02
times out URBs (a separate gate). With this PR landed, when the
chip-init gate is fixed, the descriptor will already be correct;
without it, that future work would still produce silently-dropped
frames on-air.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Collaborator
Author
|
Closing per user request — value comes only when TX works end-to-end. Will bundle the descriptor change into the eventual PR that fixes the chip-init gate as well. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Usbmon capture of the working morrownr/aircrack-ng kernel driver injecting a probe-request on the 8814 (channel 6, monitor mode) was diffed against devourer's TX descriptor, both 140-byte single-fragment frames. The first 32-byte descriptor block differed in seven fields. Update devourer's
send_packetso the descriptor is byte-identical to the kernel-driver's.MACIDRATE_ID(non-VHT)GID0x3F)SW_DEFINEDriverFixedRateflagRETRY_LIMIT_ENABLEDATA_RETRY_LIMITrtl8814au_xmit.c:267SPE_RPTDISABLE_FBBackground
The previous values came from earlier WIP comments speculating about chip behavior. Live usbmon diff against the working kernel driver shows the actual field set the chip expects.
Verification
64002885 01120800 0000003f 00010000 00003200 00000000 01000000 76a90000— byte-identical to kernel-driver's TX descriptor.insmod 8814au.ko, monitor, ch6,rmmod) and devourer started withDEVOURER_SKIP_INIT=1: 781 of 781 bulk-OUT URBs complete with status=0. With the previous descriptor, URBs also completed status=0 but on-air TX never happened (the chip silently dropped frames whose descriptor field values it didn't accept).What this does NOT fix
End-to-end 8814 TX without
DEVOURER_SKIP_INIT. Devourer's HAL init still leaves the chip in a state where EP0x02times out URBs (a separate gate, tracked in task #21). With this PR, when the chip-init gate is fixed, the descriptor will already be correct.Test plan
0bda:8813ch6DEVOURER_SKIP_INIT=1path: 781/781 bulk-OUT URB completions status=0🤖 Generated with Claude Code